I don't really see your point. AI/ML can make programs the same way as humans, which is writing a program that seems right but it very wrong and keep iterating

I don't really see your point. AI/ML can make programs the same way as humans, which is writing a program that seems right but it very wrong and keep iterating until it mostly works, most of the time.